WASHINGTON, DC - OCTOBER 30: U.S. President Joe Biden hands Vice President Kamala Harris the pen he ... [+]
Reining in abuses stemming from the misuse of artificial intelligence is now a top federal government priority. Anticompetitive behavior, the subject of antitrust laws, is among the potential AI-related harms spotlighted in an October 2023 Presidential Executive Order on AI. Antitrust will be applied to review a wide variety of business practices, including the pricing of goods and services by competitors, mergers, and actions by “dominant” firms, including tech platforms. Because commercial applications of AI are spreading like wildfire, business leaders and executives need to know what conduct enforcers are focusing on in order to avoid pitfalls that could subject them, and their firms, to major legal liability.
President Joe Biden targeted the misuse of algorithms to fix prices in last week’s State of the Union address, in which he stated that “[f]or millions of renters, we’re cracking down on big landlords who break antitrust laws by price-fixing and driving up rents.” Algorithmic-based anticompetitive behavior is already the subject of major private lawsuits, and government investigations are ramping up. Regulators should tread carefully, however, to ensure that the scrutiny does not discourage economically beneficial business applications of algorithms.
AI Basics And Competitive Risks
According to Oracle Content Strategist Michael Chen, an AI model “is both a set of selected algorithms and the data used to train those algorithms so that they can make the most accurate predictions.” The rapid growth of computing power in recent years has made AI an ever-more powerful tool used in business planning and management.
But executives should keep in mind that providing information to train AI software may create legal problems. These may include not only antitrust issues but copyright infringement, data privacy and security breaches, and anti-discrimination law violations, to name a few. In-house counsel responsible for monitoring legal risk exposure will need to work closely with firm managers to avoid such problems.
The Executive Order recognizes that AI has great potential to do good but poses major risks as well. To the extent that competition is harmed, the solution is antitrust enforcement, which the Order identifies as “addressing risks arising from concentrated control of key inputs, [and] taking steps to stop unlawful collusion and prevent dominant firms from disadvantaging competitors.”
Antitrust Enforcement And AI
The two federal antitrust enforcement agencies, the Department of Justice and FTC, are responsible for addressing the three specific competition-related concerns referred to by the AI Order: collusive conspiracies such as price-fixing, concentrated control of key inputs, and dominant firms disadvantaging competitors. DOJ challenges collusive agreements as antitrust crimes (the FTC lacks criminal enforcement authority), while both share enforcement authority to prosecute civilly anticompetitive agreements, monopolization, and anticompetitive mergers. Private parties may also seek treble damages due to anticompetitive arrangements.
Antitrust challenges to AI-related practices are in their infancy. At this time, DOJ and FTC attention have honed in on algorithm-related collusion. A recent study by antitrust scholar Satya Marar reveals the prospects and pitfalls of antitrust challenges in this area.
AI Algorithmic Collusion Cases
Collusive cartel conduct, involving secret agreements among business rivals to fix prices, is the “supreme evil of antitrust,” according to the U.S. Supreme Court.
Without an agreement, there is no criminal antitrust violation. Enforcers are concerned, however, that AI algorithms can effectuate such collusion without a specific agreement. AI algorithms “trained” on industry pricing practices may help firms predict how their competitors are likely to set prices in the future. If firms purchase the same algorithmic software, they may be able to avoid price wars and adjust instantly to price changes by competitors. This could stabilize and fix prices.
Two private civil antitrust cases involving alleged collusion through algorithms have drawn government attention.
In the RealPage case, now ongoing, renters claim that multiple landlords separately fed their nonpublic business information to the same price-setting algorithm, produced by RealPage, then used it to set rental prices. The Justice Department asserted in a November 2023 court filing that “the alleged scheme meets the legal criteria for per se unlawful price fixing.” This clear statement suggests that the DOJ will be looking for criminal prosecutions raising similar fact patterns.
The DOJ and the FTC jointly filed a legal brief in November 2023 via the McKenna Duffy v. Yardy Systems, Inc. case. That matter involves a claim by plaintiff renters that landlords unlawfully agreed “to use Yardi’s pricing algorithms to artificially inflate” multifamily rental prices.
In a March 2024 article for the FTC’s “Business Blog,” staffers Hannah Garden-Monheit and Ken Merber stressed that “an agreement to use shared pricing recommendations, lists, calculations, or algorithms can still be unlawful even where co-conspirators retain some pricing discretion or cheat on the agreement.” The plain message is that enforcers are going to be aggressive in investigating algorithmic pricing.
Algorithmic Pricing Raises Tough Questions And May Be Beneficial
The conditions under which parallel conduct involving the selection or training of AI algorithms would support a criminal conviction raises novel and difficult litigation questions. Algorithms might seem to make price-related understandings easier to reach and maintain. At the same time, as Professors Joshua Davis and Anupama Reddy point out in a report published by the University of San Francisco, algorithms may create records of the steps, inputs, and calculations underlying pricing decisions, making it easier for DOJ to identify and prosecute collusive conduct.
There is also a positive side to the business deployment of AI algorithms. As Marar (citing scholarly research) notes, algorithms may better tailor product offerings to reflect consumer preferences and allow firms to respond more quickly and efficiently to price changes. More efficient pricing allows firms to respond more effectively to consumer demand.
Regulators should weigh these benefits in setting enforcement policy to deter abuses without disincentivizing the economically desirable uses of algorithms that invigorate competitive forces.
